Comparison of different types of volatility models for NIFTY50 index

Abstract

In this thesis, volatility from dfferent models are compared using NIFTY50 index data. In the first half of thesis, we build our understanding of volatility and it’s different types. Then, we move to understanding volatility clustering using autocorrelation. To capture the effect of volatility clustering we discuss univariate models like AR, ARMA and GARCH etc. After that, we try to model volatility using Hidden Markov Switching(HMS) model and GARCH model. To know which of the model performs better, we forecast volatility using both the models and compare them using a true volatility indicator, India VIX index. In the second part of my thesis, we developed a pair trading strategy for NIFTY50 and BANKNIFTY futures using the concepts of stationarity, mean-reversion and correlation. Using HMS model to classify the market into regimes, we try to analyze the performance of our pairs strategy in binary regimes.